Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Findings

This page collects what JETCARP’s searches have found and what the checks of November 2026 made of it. It is the data behind the claims on What is new in JETCARP: every result was run again on the release engine (JETCARP 0.4.0), every spaceship was compared with the ships of simpler tilings, and the list was cut down wherever a check failed. A census of Life itself on the Archimedean, 2-uniform and 3-uniform tilings went through the same checks; its results are in their own section. The ConwayLife.com forums have not been searched yet, so a candidate here is a candidate, not a discovery; we will ask the community when the results are posted there. Corrections are welcome as issues.

Re-verification on the release engine

The finds were made with JETCARP’s predecessor, CART. Before anything was claimed, the records were run again on JETCARP 0.4.0:

RecordsCheckedPassed
Spaceships103103
Long-period oscillators (the 40 longest per tiling family)739739
Turmite highways on non-square tilings39,74439,744

Spaceships and oscillators were checked twice: by jetcarp library verify --all, and by pasting each record alone on a fresh wrapped patch and letting JETCARP’s object finder name it again after 200 generations. No period, displacement or speed changed. Every highway gave the same onset, period, cells and speed with its stored limits.

Isohedral tilings are eleven neighbour graphs

A rule that only counts neighbours sees the neighbour graph, not the shapes (see Tilings as neighbour graphs). Every isohedral tiling is combinatorially one of the eleven Laves tilings, the duals of the uniform tilings, whatever the shape of its tile, so JETCARP’s 81 isohedral types give rules only eleven different worlds in each neighbourhood. A survey of all 362 tiling and neighbourhood pairs (19 families, the 81 isohedral types and the 81 catalogued 2- and 3-uniform tilings, with touching and with edge neighbours) found:

Laves tilingIsohedral types in JETCARPNeighbour graph is that of
[36], hexagon-likeIH01–IH18, IH20 (19 types)the hexagonal grid
[44], square-like30 types from IH41 to IH76the square grid
[63], triangle-likeIH83–IH86, IH88, IH90, IH91, IH93 (8)the triangular grid
[32.4.3.4], Cairo pentagonalIH27–IH29 (3)the Cairo tiling
[33.42], prismatic pentagonalIH22–IH26 (5)itself
[34.6], floret pentagonalIH21itself
[3.6.3.6], rhombilleIH33, IH34, IH36, IH37 (4)itself
[3.4.6.4], deltoidal trihexagonalIH30–IH32 (3)itself
[3.122], triakis triangularIH38–IH40 (3)itself
[4.82], tetrakis squareIH78, IH79, IH81, IH82 (4)itself
[4.6.12], kisrhombilleIH77itself

Several drawn families are also lattices in disguise: the brick, herringbone and three-fold (lizard) tilings behave exactly as hexagons, the quadrilateral, glide and four-fold tilings as squares, and the Pythagorean tiling as 4.8.8. No 2- or 3-uniform tiling has the graph of a lattice, of Cairo or of an Archimedean tiling. The isohedral tilings page therefore presents results on an isohedral type as results on its Laves tiling.

The survey also explains an early false alarm. JETCARP’s catalogue held “spaceships” on IH46, IH54, IH62, IH72 and IH84; mapped back onto their lattices they are Conway’s glider, the lightweight and middleweight spaceships, and an unnamed ship of the triangular grid. Every ship found on an isohedral type is a known lattice object, and none is claimed.

Spaceships

How the list was cut

Of the 103 spaceship records, 35 were known objects of the square and triangular grids (every ship on an isohedral type among them) and 5 were H. S. Guttesen’s kagome glider, alone, in pairs or in corridors of kagome tiles. Fresh soup searches in November 2026 re-found Guttesen’s snub-square glider and added 16 objects. That left 72 candidates, which a second, deliberately skeptical pass and then a looser congruence test examined:

StepResult
Candidates examined72
Compounds: two smaller ships flying side by side without influencing each other10, removed
The same object as another record, with every neighbour count the samenone
The same live tiles at every generation as another candidate (congruence test)2 pairs, merged
Re-attributed to a simpler tiling by the strict map (7 from 3-uniform to 2-uniform tilings, 1 from k3-032 to 3.6.3.6)8
Re-attributed to a simpler tiling by the congruence test (5 to 3.3.3.4.4, 3 from 3-uniform to 2-uniform tilings)8
Objects that run on more than one tiling19
Distinct surviving objects60
of which single ships (6 of them contain a perturbed phase of a smaller ship)47
of which derivative: a smaller ship with a passenger that it pulls along unchanged, or a bound pair of gliders13
Flag A: most likely already known or noticed by others11
Flag B: check before claiming17
Unflagged32, of which 25 are single ships

The checks are described in the box on What is new in JETCARP. No ship depended on dead edges or on the size of the patch, no period or displacement was misreported, and no candidate is a ship of a lattice, of Cairo or of any Laves tiling.

The skeptical pass called two ships the same object only if a map between them kept every neighbour count, of the live tiles and of the empty tiles around them. The congruence test is looser: two ships are one object if, after a shift of phase, one rotation, reflection and translation carries the live tiles of one onto the live tiles of the other at every generation of a period, whatever the neighbour counts of the empty tiles around them. It merged two pairs:

  • d8692b4b and 388daffd are one 7-cell ship. On k2-003 and k3-016 the squares at the edge of its strip of squares and triangles border another row of squares and touch 10 tiles; on k3-013 and k3-019 they border a row of triangles and touch 12. The live tiles never notice: every birth and survival condition is the same, and the range only looks 16 times larger on k3-013 because counts of 11 and 12, which the ship never meets, exist only there.
  • 1a403313 and cf8efa2e are one 4-to-5-cell ship, recorded twice on the same tiling k3-019, at two positions that no symmetry of the tiling exchanges: one translation, with a shift of two generations, carries one onto the other, but the rows beyond its strip are squares on one side and triangles on the other, and swapped at the second position. Their ranges are equal.

The same test, run against every Archimedean, 2-uniform, 3-uniform and isohedral tiling (an isometric copy of each ship’s live tiles is placed on the tiling, and JETCARP runs it in a rule that the counts there allow), found that eight ships run, with the same live tiles, on a simpler tiling than the one they were claimed on: d8692b4b, b86608f6, d36fd2c3, 9f3f8ff8 and 1a403313, found in strips of squares and triangles inside 2- and 3-uniform tilings, run on the Archimedean elongated triangular tiling 3.3.3.4.4; 9c87573c (from k3-003) runs on k2-012; and b79fe209 and b3726ba1 (from k3-015 and k3-021) run on k2-015. They are listed on the simpler tiling, and so, by the same rules as the others, d8692b4b and 1a403313 are now flagged: small, common in their searches and wide-range on an Archimedean tiling.

Flag A marks ships that are small, turn up often in soups and run in many rules on an Archimedean tiling (78d55497, bde55795, 30594bf3, 3aaf02c0, 781cc2ef, 1df4b31f, d8692b4b, 1a403313), and the ships built on Guttesen’s gliders (c696c65e, a5e9ba80 in Life on 3.3.4.3.4, and e08df0a8). Others may well have seen these; we found no report, and do not write “to our knowledge” for them. Flag B marks ships that run in a well-known rule on an Archimedean tiling (HighLife, Pedestrian Life, or the B2/S34 rule of the 4.8.8 hunts, which is Hexagonal Life without its H), that contain a phase of one of Guttesen’s gliders, or that were very common in their search. Four more joined Flag B after a second search of the ConwayLife forums found EricG’s 2013 exploration of the Archimedean tilings (EricG 2013): 7f22824b and 504fdf46 run in B2/S2 on 4.8.8, where EricG found “many of the B2 rules” to have spaceships, and 95528302 and 18746b1c run in B26 rules on 3.12.12, next to the B2/S2 and B26/S3569 rules in which EricG reported many ships. EricG did not list the ships, and the pattern files are no longer online, so these four wait for a comparison with them. The claimed ships on 3.4.6.4, 3.3.3.4.4 and 3.3.3.3.6 run in none of the rules EricG named (see What is new in JETCARP).

Life itself (B3/S23), the rule Guttesen searched, deserves a separate word. All eleven of these candidates that run in Life turned out to be made of Guttesen’s gliders: on 3.6.3.6 every one is a pair of glider g, so no new single Life ship on the kagome tiling survives; on 3.3.4.3.4 one is a pair of the snub-square glider, one is a bound pair of it (c696c65e, period 16) and one contains a perturbed copy of it (a5e9ba80, period 24). But the searches behind these candidates ran Life on only a few tilings, so they say little about Life elsewhere. A later census of Life on all 92 Archimedean, 2-uniform and 3-uniform tilings found Life spaceships on 28 of them, 24 of those 2- or 3-uniform; see Life on the Archimedean and k-uniform tilings.

The 60 objects

Sorted by the simplest tiling each one runs on: Archimedean, then 2-uniform, then 3-uniform. Runs on lists every tiling where the ship was catalogued or where JETCARP confirmed it as one ship whose live tiles evolve exactly as the catalogued ship’s: either a map that keeps all neighbour counts carries it there, or an isometric copy of it runs in some rule while the empty tiles around it have other neighbour counts. Merged objects give the second id in brackets. d per period is the displacement in lattice steps, in the lattice basis of the patch it was catalogued on; on the tilings with a hexagonal lattice the two basis vectors are 60° apart, so (1, −1) is as short a step as (1, 0), and direction words such as “diagonal” would mislead. Smallest rule is the rule with the fewest birth and survival conditions in which the ship runs; it runs identically in every rule between that one and the largest, and the count says how many that is; where two counts are given, the count depends on the tiling (each card lists them), because a tiling with more neighbours per tile leaves more counts that the ship never meets free. Seen is how often the search met it, which depends on how long that search ran. Each id links to the object’s card in the gallery below, with a clip of it moving.

idruns on (simplest first)pd per periodcellssmallest rule (rules in range)seenkindflag
19d091ca34.8.82(0, 1)5–9B2/S3 (512)270singleB
2ca2f31b04.8.82(0, −1)14–16B2/S2 (256)8derivative: 78d55497 with a tag-along
378d554974.8.82(0, −1)5B2/S2 (512)8,034singleA
42f5d31824.8.84(−2, 0)9–10B2/S2 (256)32derivative: bde55795 with a tag-along
5298db0cf4.8.84(0, −2)9–14B2/S2 (64)5derivative: 78d55497 with a tag-along
6984c5dd54.8.84(0, −2)7–10B2/S2 (256)92derivative: 78d55497 with a tag-along
787759c784.8.84(0, −2)14–17B2/S2 (256)2derivative: 78d55497 with a tag-along
8feae01724.8.84(0, −2)10–15B2/S3 (64)54derivative: 9d091ca3 with a tag-alongB
9bde557954.8.84(0, −2)5–6B2/S2 (256)406singleA
10c45a97a14.8.84(0, −2)8–9B2/S2 (256)3derivative: 7f22824b with a tag-along
117f22824b4.8.84(0, −2)7B2/S2 (512)9single, contains a phase of 78d55497, bde55795B
12504fdf464.8.86(0, 2)7–13B2/S2 (128)11singleB
13908dc7a04.8.812(6, 0)12–18B2/S34 (64)2derivative: 9d091ca3 with a tag-alongB
1487dfbfaf4.8.816(0, 8)19–26B2/S34 (16)1derivative: 9d091ca3 with a tag-alongB
1571f3b4b34.8.816(0, 8)14–20B2/S34 (32)95derivative: 9d091ca3 with a tag-alongB
16768974844.8.816(0, 8)14–28B2/S34 (32)271singleB
17a333894b4.8.824(12, 0)14–21B2/S3 (128)1singleB
18955283023.12.122(0, −1)6–9B26/S2 (32,768)39single, contains a phase of 30594bf3B
1918746b1c3.12.122(0, −1)7–9B26/S2 (16,384)6single, contains a phase of 95528302, 30594bf3B
2030594bf33.12.122(0, 1)4–5B2/S2 (65,536)1,742singleA
213aaf02c03.12.126(−2, 4)5–8B2/S2 (32,768)1,585singleA
2225ba593a3.12.128(4, 0)12–23B26/S26 (1,024)1derivative: 95528302 with a tag-along
23781cc2ef4.6.124(0, 1)5–6B2/S2 (65,536)373singleA
2457a0326c3.4.6.442(0, 3)7–15B3/S235 (1,024)61single
25d8692b4b (= 388daffd)3.3.3.4.4, k2-003, k2-004, k3-010, k3-013, k3-016, k3-019, k3-051, k3-0524(−1, 0)7B45/S34 (256 to 4,096)447singleA
26b86608f63.3.3.4.4, k2-003, k2-004, k3-010, k3-013, k3-016, k3-019, k3-051, k3-0524(−1, 0)12–16B456/S347 (256 to 4,096)7single
271a403313 (= cf8efa2e)3.3.3.4.4, k2-003, k2-004, k3-010, k3-013, k3-016, k3-019, k3-051, k3-0525(−1, 0)4–5B3/S2 (8,192 to 131,072)406singleA
283e30b9ff3.3.3.4.4, k2-003, k2-004, k3-010, k3-013, k3-016, k3-019, k3-051, k3-0528(2, 0)13–17B46/S3458 (64 to 1,024)2single
29d36fd2c33.3.3.4.4, k2-003, k2-004, k3-010, k3-013, k3-016, k3-019, k3-051, k3-05218(−1, 0)10–14B4567/S3478 (32 to 512)45single
309f3f8ff83.3.3.4.4, k2-003, k2-004, k3-010, k3-013, k3-016, k3-019, k3-051, k3-05226(1, 0)17–26B45679/S3478 (16 to 256)1single
31d3cd39513.3.3.3.627(−1, 1)4–18B37/S2,5,6,10 (1,048,576)30single
32c696c65e3.3.4.3.416(0, −4)15–33B3/S23 (256)1derivative: a bound pair of Guttesen’s snub-square gliderA
33a5e9ba803.3.4.3.424(0, 6)17–50B3/S23 (128)1single, contains a phase of Guttesen’s snub-square gliderA
348f790b163.6.3.6, k3-030, k3-032, k3-0394(−1, 0)8–9B36/S23 (8,192)423single, contains a phase of Guttesen’s gB
351df4b31f3.6.3.6, k3-030, k3-032, k3-0394(−1, 0)7–8B3/S345 (16,384)1,347singleA
36e08df0a83.6.3.64(−1, 0)10–13B36/S23 (4,096)2derivative: Guttesen’s g with a tag-alongA
3748e9e2673.6.3.612(0, 3)5–12B38/S23 (2,048)595single, contains a phase of Guttesen’s gB
382e4467c0k2-004, k2-003, k3-010, k3-013, k3-016, k3-0193(−1, 0)4–6B3/S35 (2,048 to 32,768)19single
396469dbc4k2-004, k2-003, k3-010, k3-0165(−1, 0)5–9B3/S356 (512)4single
40e71e1136k2-006, k3-028, k3-029, k3-03010(1, 0)4–9B36/S23 (4,096)43singleB
419bebe7eak2-006, k3-02816(1, 0)8–11B3/S1346 (2,048)13single
423fcb35b1k2-00646(−1, 2)5–17B38/S345 (512)26single
43af899881k2-007, k3-028, k3-031, k3-0324(−1, 0)6–13B3/S13468 (4,096)17single
443021704ck2-012, k3-0036(−1, 0)7B45/S34 (1,024 to 262,144)110single
459c87573ck2-012, k3-00310(1, 0)4–10B3/S2456 (512 to 131,072)254single
46a5564155k2-012, k3-00314(−1, 0)11–17B456/S3468 (128 to 32,768)11single
479d1ccd9ck2-0136(−1, 0)9–20B2,10/S1 (4,294,967,296)44single
48e1067c81k2-01317(1, −1)2–20B2,8,13/S1 (2,147,483,648)275single
4949a92bbek2-014, k3-014, k3-020, k3-0514(−2, 0)5–6B3/S2 (4,096 to 16,384)179single
50b79fe209k2-015, k3-015, k3-021, k3-05232(1, 0)11–16B4,5,6,7,8,11/S56 (16)30single
51b3726ba1k2-015, k3-015, k3-021, k3-05244(−1, 0)11–16B4,5,6,7,8,10/S56 (16)35single
522afbf92bk2-01717(0, −1)3–13B3/S25 (8,192)949singleB
537ff20168k3-00916(−1, 0)4–10B3478/S23 (134,217,728)466single
54d5600a12k3-02212(−1, −1)9–11B3/S3457 (1,024)1,158singleB
55ed99f9a8k3-0323(1, 0)10–13B3/S3457 (1,024)2single
566c774748k3-03853(−1, 0)4–24B3/S136 (2,048)4single
5717453dabk3-04537(1, 0)6–23B4,5,9,11/S2,3,5,8,9,14 (4,096)8single
589242617ek3-05326(0, 1)3–7B36/S25 (16,384)642single
59ba6acb83k3-05812(0, 1)6–10B37/S25 (16,777,216)1,096singleB
602c3cee67k3-05822(−1, 1)4–12B37/S25 (4,194,304)359single

Two notes on the table. The two k2-013 ships run identically in about 232 and 231 rules because they use very few neighbour counts, although that tiling’s dodecagons touch up to 24 tiles: the ships are real, but their mechanism is simple. And a derivative ship is kept as a distinct object, as Life catalogues keep tag-alongs, but it is not counted as a separate discovery.

Not claimed

  • Compounds. Ten catalogue entries are two ships that fly side by side and never influence each other; JETCARP’s object finder stored each pair as one ship. Eight are two of Guttesen’s kagome glider g (76704620, 3f3cce70, 6327b0f4, 9cd171f7, 98597b1c, 8054f744, fe08fb1a, cca8e8f5), one is two of Guttesen’s snub-square glider (5186d210) and one is two copies of 48e9e267 (83f8eebe).
  • Guttesen’s gliders. Glider g on 3.6.3.6 and the period-8 glider on 3.3.4.3.4, both in Life, were found by JETCARP’s searches independently; they are Guttesen’s, first posted as Code Golf answers in November 2024 (Guttesen 2024; Guttesen 2026). EricG had already reported spaceships in Life on both tilings in 2013, without describing them (EricG 2013). JETCARP’s 4-to-5-cell kagome ship is exactly Guttesen’s Fig. 4 glider (the same cells in one phase), and its snub-square ship matches Fig. 5 by eye; a cell-for-cell check awaits Guttesen’s published configurations. Guttesen’s large kagome glider G (period 40) was not found. Glider g also runs in corridors of kagome tiles inside the 3-uniform tilings k3-030, k3-032 and k3-039; the snub-square glider runs only on 3.3.4.3.4.
  • Lattice objects. Every ship on an isohedral type, and the unnamed ships of the square and triangular grids found along the way, belong to those grids.

Every one of the 60 objects, in the order of the table, grouped by the verdicts of the skeptical pass and the congruence test. Each clip is a fixed window that the ship crosses, on the first tiling of its runs on list, in the rule the search found it in, with the tile outlines drawn. The Tiling line lists every tiling on which the ship’s live tiles evolve the same way; where the empty tiles around them have other neighbour counts, the range of rules differs, and the Rules line gives each. Watch each ship come back to the same shape one period later, moved along by its displacement; the clip then starts again. Speeds are in lattice steps per period, as in the table; where a ship moves along both lattice vectors, the card gives its displacement. Slow ships are filmed every few generations, and each card says which generations its clip shows. With reduced motion, the clip is replaced by a still of a few of its positions, older ones fainter. The scenes are in docs/scenes/findings/, one per object, and docs/tools/render-animations.sh findings/ makes the clips again.

Single ships with no flag (25)

The 25 single ships that carry no flag. These are the only candidates that What is new in JETCARP claims, to our knowledge.

Derivatives with no flag (7)

A smaller ship that pulls a passenger along unchanged. Each is kept as a distinct object, with its core linked, but is not counted as a discovery. Six more derivatives are flagged and shown below.

Flag A: most likely already known (11)

Small ships that turn up often in soups and run in many rules on an Archimedean tiling, and the ships built on Guttesen’s gliders. We found no earlier report of them, but others may well have seen them, so they are not claimed.

Flag B: check before claiming (17)

Ships that run in a well-known rule on an Archimedean tiling, contain a phase of one of Guttesen’s gliders, were very common in their search, or run on 4.8.8 or 3.12.12 in the rules where EricG reported many spaceships in 2013 without listing them. We found no earlier report of them, but others may well have seen them, so they are not claimed.

Life on the Archimedean and k-uniform tilings

The candidates above came from searches that ran Conway’s Life itself on only a few tilings, so what they say about Life is incomplete. In November 2026 JETCARP therefore ran a census of Life (B3/S23, with cells that share an edge or a vertex as neighbours) on 92 tilings: the 11 Archimedean tilings, which include the square, hexagonal and triangular grids, the 20 2-uniform tilings and the 61 3-uniform tilings of the k-uniform catalogue. On each it ran 50,000 random soups of 16 by 16 unit cells, each cell live with probability one half, on the unbounded plane until the soup settled, and catalogued every object that was left and every spaceship that flew away (jetcarp search hunt, the job labelled life-k123).

Life has spaceships on 28 of the 92 tilings: four Archimedean tilings (the square grid, 3.6.3.6, 3.3.3.4.4 and 3.3.4.3.4), five of the 2-uniform tilings and 19 of the 3-uniform ones. On the other 64, no soup produced one. That is not a proof that they have none: a ship that turns up in fewer than about one soup in 50,000 could easily have been missed, and so would any ship of period above 64, which the census did not look for.

TilingSoupsShips seenDistinct objectsOf which listed here
square50,00095,22764 known, 2 compound
hexagonal50,000none
triangular50,000none
4.8.850,000none
3.6.3.650,000149,878131 known, 12 compound
3.3.3.4.450,0004,60521 known, 1 compound
3.3.4.3.450,000118,322211 known, 1 flagged, 5 derivative, 14 compound
3.3.3.3.650,000none
3.4.6.450,000none
3.12.1250,000none
4.6.1250,000none
k2-00150,000none
k2-00250,000none
k2-00350,00057721 single, 1 flagged
k2-00450,000none
k2-00550,000none
k2-00650,000none
k2-00750,0005,77331 known, 1 flagged, 1 compound
k2-00850,000none
k2-00950,000none
k2-01050,000none
k2-01150,00033211 flagged
k2-01250,000111 single
k2-01350,000none
k2-01450,000none
k2-01550,000none
k2-01650,000none
k2-01750,00067,75331 flagged, 1 derivative, 1 compound
k2-01850,000none
k2-01950,000none
k2-02050,000none
k3-00150,000none
k3-00250,00056021 single; 1 listed on k2-011
k3-00350,000none
k3-00450,000none
k3-00550,000none
k3-00650,000none
k3-00750,000none
k3-00850,000none
k3-00950,000none
k3-01050,00038442 single; 1 listed on square; 1 listed on k2-003
k3-01150,0002644 single
k3-01250,000none
k3-01350,0004,24821 compound; 1 listed on 3.3.3.4.4
k3-01450,000none
k3-01550,000none
k3-01650,00037911 listed on k2-003
k3-01750,000none
k3-01850,000none
k3-01950,0004,74411 listed on 3.3.3.4.4
k3-02050,000none
k3-02150,000none
k3-02250,000none
k3-02350,000none
k3-02450,000none
k3-02550,000none
k3-02650,000none
k3-02750,000none
k3-02850,000none
k3-02950,000none
k3-03050,00035,75097 compound; 2 listed on 3.6.3.6
k3-03150,000none
k3-03250,00035,643119 compound; 2 listed on 3.6.3.6
k3-03350,000none
k3-03450,000none
k3-03550,000none
k3-03650,000none
k3-03750,000none
k3-03850,000none
k3-03950,00047,294131 flagged, 2 derivative, 8 compound; 2 listed on 3.6.3.6
k3-04050,0003,67531 single, 1 flagged, 1 compound
k3-04150,00015011 single
k3-04250,000none
k3-04350,000none
k3-04450,00088,580451 single, 1 flagged, 3 derivative, 40 compound
k3-04550,0007,64431 flagged, 2 compound
k3-04650,0003,79421 derivative; 1 listed on k3-045
k3-04750,000none
k3-04850,000none
k3-04950,000none
k3-05050,000none
k3-05150,000none
k3-05250,000none
k3-05350,00070,91411 listed on k2-017
k3-05450,000none
k3-05550,0002661413 single, 1 derivative
k3-05650,000none
k3-05750,0006,61021 single, 1 flagged
k3-05850,000none
k3-05950,00081,623121 flagged, 1 derivative, 10 compound
k3-06050,000none
k3-06150,00034,070171 flagged, 1 derivative, 15 compound

Ships seen counts every ship the census met, so a soup that sent out three gliders counts three. Distinct objects counts the objects among them after the checks below; of which listed here counts those listed on this tiling, since an object that also runs on a simpler tiling is listed there.

How the census ships were checked

The census catalogued 198 distinct spaceship records. Each went through the same checks as the candidates above:

  1. JETCARP ran it alone on a large wrapped patch, on a smaller one and with dead edges, and its object finder saw one spaceship each time, with the recorded period and displacement; a scratch catalog recomputed the same range of rules. The object finder cannot do this for a ship whose parts lie too far apart to be grouped (it sees the parts of a bound pair as two ships, and the pieces of one large k3-055 ship, 5f371c4c, as debris); for those, JETCARP’s engine was checked to repeat the ship, translated, for 12 periods.
  2. Every phase was split into groups and each group run alone, to find compounds: two or more smaller ships flying together without influencing each other.
  3. Every smaller ship of the same tiling was searched for inside it, to find derivatives: a smaller ship that runs inside it unchanged and pulls a passenger along, or two ships that are bound together.
  4. Every pair of records, and every record against the candidates above and Guttesen’s gliders, was tested for congruence: the same live tiles at every generation, after one rotation, reflection and translation and a shift of phase, whatever the neighbour counts of the empty tiles around them. A congruent pair is one object.
  5. Each object’s live tiles were placed on all 181 tilings of the earlier checks (Archimedean, 2-uniform, 3-uniform and isohedral), and JETCARP ran every placement that Life allows. An object is listed on the simplest tiling where it runs in Life: Archimedean before 2-uniform before 3-uniform.
  6. Each object was compared with the known Life ships: Conway’s glider and spaceships, Guttesen’s two gliders, the 3.3.3.4.4 ship that EricG reported, and Büttner’s period-20 ship. The last two could only be compared with their published descriptions, since their patterns are not available.
VerdictObjects
Spaceship records in the census198
Records added from the 500,000-soup hunt on k2-0071
Distinct objects (records congruent to one another merged)185
Compounds: two or more smaller ships flying together124
Derivatives: a smaller ship with a passenger, or a bound pair15
Known objects, credited below8
Ships of a simpler tiling0
Single ships, flagged12
Single ships with no flag26

Two kinds of verdict need a word. A ship found in a corridor of a simpler tiling inside a 3-uniform tiling is the same object as the simpler tiling’s ship, and every such ship in this census was also found on the simpler tiling itself: Guttesen’s glider g in the kagome corridors of k3-030, k3-032 and k3-039, the lightweight spaceship in the strips of squares of k3-010, the probable EricG ship in the strips of squares and triangles of k3-013 and k3-019, and three ships that run on a 2-uniform tiling and in strips of 3-uniform ones. So no object had to be moved to a simpler tiling. And a flagged ship passed every check, but is so common (in at least one soup in a hundred on a tiling where the census met it) that anyone who ran Life there would see it at once, or lies on a tiling where others have run Life before (3.3.4.3.4 and k2-007). For these the question is not whether the ship is new, but whether anyone has run Life on that tiling. We know of no one who has run Life on a 3-uniform tiling, or on a 2-uniform tiling other than k2-007; other rules have been run on k2-001, where the census found no Life ship (see What is new in JETCARP).

Most census records are not new objects. JETCARP’s object finder stores ships that fly side by side as one, so the census lists 124 compounds, nearly all pairs (and a few triples) of the commonest ship of their tiling: 40 of the period-31 ship of k3-044, 29 of Guttesen’s glider g on 3.6.3.6 and in kagome corridors, 15 of the period-22 ship of k3-061, 14 of the snub-square glider, 10 of the period-30 ship of k3-059, 7 of the period-20 ship of k3-039 (three of them with a glider g), and a few more, among them two pairs of Conway’s glider and two of the probable EricG ship. The 15 derivatives are 14 bound pairs (five of them made of the snub-square glider) and one large k3-055 ship that carries a smaller k3-055 ship unchanged and drags debris behind it. Six single ships contain a perturbed phase of a smaller ship without being built from it, as Life’s own heavyweight spaceship contains a phase of the lightweight one: the period-24 ship on 3.3.4.3.4 (Guttesen’s glider), the period-20 ship of k3-039 (glider g), the k3-010 ship 25c5dc4a (the lightweight spaceship, which runs unchanged in the strips of squares of k3-010), and three ships of k3-002 and k3-011 that each contain a phase of a smaller ship of their own tiling.

The large ships of k3-055 deserve a caution. The census found 14 of them, all moving at c/4, with periods of 20, 32 and 64 and up to 169 cells, each seen in only 1 to 185 soups. No test splits them into independent parts, and only one of them carries a smaller one unchanged, so 13 are listed as single ships; but they look alike, and a closer look may well find that they share a common engine.

The Life objects

The known objects and every single ship that passed the checks, flagged or not, in the order of the gallery below. Seen counts sightings on every tiling where the census met the object, with the number of soups in brackets; rules in range gives the smallest rule (always B3/S23) and how many rules the ship runs in unchanged. Each id links to its card.

idtiling (also runs on)pd per periodcellsrules in rangeseen (soups)verdict
177364004square (25 isohedral and drawn tilings with the square grid’s neighbour graph)4(−2, 0)11–15B3/S23 (512)45 (45)known: the middleweight spaceship (MWSS); contains a phase of the LWSS
25a080db3square (25 isohedral and drawn tilings with the square grid’s neighbour graph)4(−1, −1)5B3/S23 (512)95,041 (29,732)known: Conway’s glider
36dd55e69square (k3-010, 25 isohedral and drawn tilings with the square grid’s neighbour graph)4(−2, 0)9–12B3/S23 (512)135 (133)known: the lightweight spaceship (LWSS)
41376072esquare (25 isohedral and drawn tilings with the square grid’s neighbour graph)4(−2, 0)13–18B3/S23 (512)11 (11)known: the heavyweight spaceship (HWSS); contains a phase of the MWSS and the LWSS
547fc3f433.6.3.6 (k3-030, k3-032, k3-039)4(1, 0)4–5B3/S23 (8,192)245,860 (114,606)known: Guttesen’s kagome glider g
6232055ff3.3.3.4.4 (k3-013, k3-019)2(−1, 0)8–12B3/S23 (8,192)13,595 (13,052)known: EricG’s 3.3.3.4.4 ship
7c574f8e33.3.4.3.48(0, −2)8–15B3/S23 (8,192)118,271 (43,179)known: Guttesen’s snub-square glider
8fabe952bk2-00720(1, 0)6–20B3/S23 (8,192)5,772 (5,537)known: Büttner’s period-20 ship
90e3664c5k2-00313(−2, 0)12–44B3/S23 (2,048)13 (13)single
10d02b5ce8k2-0123(−1, 0)10–14B3/S23 (32,768)1 (1)single
11ca75ccd6k3-0023(−1, 0)14–16B3/S23 (1,024)10 (10)single; contains a phase of f3d41f14
1225c5dc4ak3-0104(−2, 0)11–15B3/S23 (2,048)2 (2)single; contains a phase of the LWSS
1341eb5c62k3-01038(4, 0)8–24B3/S23 (2,048)23 (23)single
148488e7e9k3-0114(1, 0)11–15B3/S23 (8,192)8 (8)single; contains a phase of d997330c
15d997330ck3-0114(−1, 0)8–12B3/S23 (8,192)9 (9)single
164f66f17ek3-0114(−1, 0)11–18B3/S23 (8,192)1 (1)single; contains a phase of 8488e7e9 and d997330c
17f87a9ed1k3-0119(−1, 0)9–16B3/S23 (8,192)8 (8)single
18b6a6a47ak3-0408(2, 0)8–12B3/S23 (32,768)13 (13)single
19148310e3k3-0414(1, 0)12–19B3/S23 (32,768)150 (150)single
20f6567467k3-0448(1, 0)8–18B3/S23 (131,072)5 (5)single
21139723e0k3-05520(5, 0)35–57B3/S23 (8,192)185 (185)single
2265acb120k3-05532(8, 0)17–65B3/S23 (8,192)21 (21)single
235f371c4ck3-05532(−8, 0)28–66B3/S23 (8,192)1 (1)single
245912456dk3-05532(−8, 0)73–121B3/S23 (8,192)1 (1)single
25fbad7fbfk3-05532(8, 0)53–105B3/S23 (8,192)7 (7)single
26ad91d936k3-05532(8, 0)54–94B3/S23 (8,192)1 (1)single
278130b388k3-05532(−8, 0)52–96B3/S23 (8,192)8 (8)single
28b86bd684k3-05532(8, 0)39–73B3/S23 (8,192)3 (3)single
2950ecb3d5k3-05532(8, 0)38–104B3/S23 (8,192)4 (4)single
30d878334ek3-05564(16, 0)63–148B3/S23 (8,192)6 (6)single
3192b7544ak3-05564(16, 0)75–149B3/S23 (8,192)5 (5)single
32126db117k3-05564(16, 0)28–84B3/S23 (8,192)15 (15)single
331ccdebc8k3-05564(16, 0)103–169B3/S23 (8,192)4 (4)single
34f2841d4ck3-05712(0, −1)13–24B3/S23 (8,192)57 (57)single
35a5e9ba803.3.4.3.424(0, 6)17–50B3/S23 (8,192)3 (3)flagged; contains a phase of Guttesen’s snub-square glider
36f89d8f11k2-003 (k3-010, k3-016)16(5, 0)8–16B3/S23 (2,048)1,295 (1,292)flagged
3755db03c8k2-0075(0, −1)16–22B3/S23 (8,192)1 (1)flagged
38c1d7a48ek2-011 (k3-002)3(1, 0)9–11B3/S23 (8,192)882 (878)flagged
3916bc37c9k2-017 (k3-053)4(−1, 0)6–9B3/S23 (8,192)138,656 (77,024)flagged
40020ec7b7k3-03920(−5, 0)7–21B3/S23 (8,192)22,100 (17,978)flagged; contains a phase of glider g
41dab6fb01k3-04024(4, 0)7–23B3/S23 (32,768)3,661 (3,567)flagged
42e1fcfa73k3-04431(−2, 0)3–18B3/S23 (131,072)88,492 (42,183)flagged
433792824dk3-045 (k3-046)11(−2, 0)8–16B3/S23 (131,072)11,435 (10,811)flagged
441e6832e3k3-0578(1, 0)9–14B3/S23 (8,192)6,553 (6,141)flagged
4593a72ebdk3-05930(1, 1)7–23B3/S23 (524,288)81,611 (41,207)flagged
4658e6d5fck3-06122(1, 1)5–10B3/S23 (524,288)34,039 (25,346)flagged

Each clip is a window that the ship crosses, in Life, on the tiling it is listed on, with the tile outlines drawn; with reduced motion it is replaced by a still of a few of its positions. The scenes are in docs/scenes/life-census/, one per object, and docs/tools/render-animations.sh life-census/ makes the clips again.

Known objects (8)

Conway’s glider and spaceships, which run on the square grid and on every isohedral tiling with its neighbour graph; Guttesen’s two gliders; the ship that is probably EricG’s on 3.3.3.4.4; and the ship that is very likely Büttner’s on k2-007.

Single ships with no flag (26)

Single ships that carry no flag. To our knowledge, none of them has been reported before.

Flagged single ships (12)

Single ships that passed every check but are common on their tiling, or lie on a tiling where others have run Life. We found no earlier report of any of them.

Credits for Life on these tilings

Conway’s glider and the light-, middle- and heavyweight spaceships are the classic ships of Life (Gardner 1970; Berlekamp, Conway and Guy 1982). EricG ran Life-like rules on the eight mixed Archimedean tilings in 2013 and reported Life spaceships on 3.6.3.6, 3.3.4.3.4 and 3.3.3.4.4 (EricG 2013). Martin Büttner found a period-20 Life ship on a tiling of hexagons, squares and triangles in 2014 (Code Golf 2014). H. S. Guttesen described the kagome glider g and the snub-square glider, first in Code Golf answers in 2024 (Guttesen 2024; Guttesen 2026), and a YouTube channel showed Life gliders on the snub square tiling in May 2024 (Tessellation Automata 2024). The 2- and 3-uniform tilings are Galebach’s catalogue (Galebach: uniform tilings).

The hat glider is Edmond’s glider B

The hat glider that JETCARP’s reimplementation of James Edmond’s evolutionary search found is Edmond’s hat glider B (Edmond 2026) written in another table. Rename B’s states (1 as 3, 2 as 1, 3 as 2) and the two tables agree on every configuration the glider meets after its first generation; run side by side, they give the same states every generation, for 180 generations on a level-5 patch and 1,467 generations on a level-7 patch. The glider has one head and four to six wing tiles, its eight shapes are all among B’s ten, and it moves one ring of tiles every two generations. JETCARP’s table has one extra row and its seed launches the glider on a different lane, which is why it first looked different. The glider is Edmond’s. What JETCARP adds is a long run: on a sliding window that follows the glider, it travelled 1,000,000 rings of tiles at a steady 2.14 units per generation (Edmond’s certified minimum speed is 2.1397), and Edmond’s own certificates prove it immortal. See Priority rules.

Still to do

  • Ask on the ConwayLife.com forums whether any of the candidates, above all the flagged ones, have been seen before. Our prior-art search could not read the forums directly.
  • Compare the census’s 3.3.3.4.4 ship with EricG’s 2013 patterns, and its period-20 k2-007 ship with Büttner’s 2014 ship, cell for cell, if those patterns can be found.
  • Look more closely at the large Life ships of k3-055, which may share a common engine.
  • Run the Life census on more soups, on the k-uniform tilings with k from 4 to 6, and with ships of periods above 64, which the census did not look for.
  • Compare the kagome and snub-square ships with Guttesen’s configurations cell for cell, once they are published.
  • Extend the equivalence survey to the k-uniform tilings with k from 4 to 6, and the ship comparison to aperiodic tilings. Neither can turn a claimed ship into a lattice ship, but either could add “also runs on” entries.
  • Re-run the key searches just before publication, since several closely related works appeared in August to October 2026.